Post on 09-Jul-2020
SENSORS INTEGRATION FOR SMARTPHONE NAVIGATION: PERFORMANCES
AND FUTURE CHALLENGES
I. Aicardi a, P. Dabove a, A. Lingua a, M. Piras a
a Politecnico di Torino, DIATI – Department of Environmental, Land and Infrastructure Engineering, Torino, Italy –
(irene.aicardi – paolo.dabove – andrea.lingua – marco.piras)@polito.it
Commission III, Symposium 2014
KEY WORDS: Calibration, Inertial sensor, Localization, Navigation, SIFT, Smartphone
ABSTRACT:
Nowadays the modern smartphones include several sensors which are usually adopted in geomatic application, as digital camera,
GNSS (Global Navigation Satellite System) receivers, inertial platform, RFID and Wi-Fi systems.
In this paper the authors would like to testing the performances of internal sensors (Inertial Measurement Unit, IMU) of three
modern smartphones (Samsung GalaxyS4, Samsung GalaxyS5 and iPhone4) compared to external mass-market IMU platform in
order to verify their accuracy levels, in terms of positioning. Moreover, the Image Based Navigation (IBN) approach is also
investigated: this approach can be very useful in hard-urban environment or for indoor positioning, as alternative to GNSS
positioning.
IBN allows to obtain a sub-metrical accuracy, but a special database of georeferenced images (Image DataBase, IDB) is needed,
moreover it is necessary to use dedicated algorithm to resizing the images which are collected by smartphone, in order to share it
with the server where is stored the IDB. Moreover, it is necessary to characterize smartphone camera lens in terms of focal length
and lens distortions.
The authors have developed an innovative method with respect to those available today, which has been tested in a covered area,
adopting a special support where all sensors under testing have been installed. Geomatic instrument have been used to define the
reference trajectory, with purpose to compare this one, with the path obtained with IBN solution. First results leads to have an
horizontal and vertical accuracies better than 60 cm, respect to the reference trajectories. IBN method, sensors, test and result will be
described in the paper.
1. INTRODUCTION
During the last years, the possibility to know our position has
becoming more and more important. The users want to use their
devices to get this information (for example for location based
services) and to share it to other people.
Nowadays, the most common device are based on smartphones
technology, where several sensors are installed, as GNSS
(Global Navigation Satellite System) receiver, video-cameras,
pressure sensor, inertial Measurement Unit (IMU) with
accelerometers, gyroscopes and magnetometers. All of them
contribute to define a 3D position, but they could be used for
Geomatics purpose also.
As known, in greater part of the cases GNSS receiver is adopted
for outdoor positioning, where it allows to reach a good level of
precision, but somewhere the signal is too noisy or not available
(i.e. indoor or urban canyons) and GNSS positioning is not
allowed. The future trend of positioning is to have a seamless
solution, that means to have a continuous and stable localization
everywhere, from outdoor to indoor scenario.
In the last years, many research groups are working to study
different solutions to bridge this gap, as alternative to GNSS
positioning, using different kind of sensors.
Summarizing the main available technologies, it is possible to
divide them considering the different fields:
Wi-Fi: this technology is especially dedicated for
indoor environments in a transmission range between
30-200 m. In particular, the positioning is based on
the time-of-flight range measurements observed from
several base stations applying a triangulation. This
procedure brings to have good performance, but it
suffers from outliers, signal coverage and depends to
Access Points (AP) and geometric distribution (DoP)
(Schatzberg et al., 2014; Hatami et al., 2005; Werner
et al., 2014);
pedestrian tracking system: this procedure is based on
the use of a pedometer, that is now also available in
the modern smartphone (Yunye et al., 2011;
Woodman and Harle, 2008; Shin et al., 2014). The
possibility to adopting external sensors as low-cost
IMU-MEMS (Micro Electro-Mechanical Systems),
which was directly installed on the foot of the user has
been investigated (Yuan et al., 2014);
Bluetooth: the technology is based on the use of
Bluetooth Low Energy by adding direction finding
capability (Kallioka, 2011). There is also another
approach based on a range-free localization system
using commercial smartphones with Bluetooth
capabilities. In the range-free localization system,
each smartphone periodically scans nearby Bluetooth
enabled devices and sends the results to the
localization server. This server collects the scanning
results into a short period and find their locations
using range-free algorithms (Lee et al. 2014).
Inertial sensor navigation (Woodman 2007): another
alternative approach is to use the accelerometers
(Kunze et al. 2009), the gyroscope and magnetometers
in the pedestrian navigation in order to correct the
drifts (Afzal et al. 2011) and also in the use of
barometer sensor to identify the movements (Frank et
al. 2014), with purpose to realize a navigation.
Positioning and navigation are usually estimated adopting an
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-3, 2014ISPRS Technical Commission III Symposium, 5 – 7 September 2014, Zurich, Switzerland
This contribution has been peer-reviewed.doi:10.5194/isprsarchives-XL-3-9-2014 9
integration between the techniques described above, in order to
improve the accuracy and the precision. Nowadays, the range of
precision for positioning with these techniques is between 50 m
to 60 cm. The accuracy and precision level mainly depends to
the data processing, especially in the most of cases the
positioning are estimated using the Kalman filter. It is very
useful when inertial sensor are adopted, with purpose to predict
the bearing of the object.
The proposal of the authors is based on the integration between
images and inertial sensors which are included in the modern
smartphones, developing an innovative integration approach.
A positioning procedure based only on the use of the images
have been tested, obtaining an accuracy equal around to 30 cm.
After that, the inertial sensors have been considered in order to
estimate an integrated solution of navigation.
2. THE IMAGE-BASED NAVIGATION APPROACH
The Image-Based Navigation (IBN) procedure is an approach
devotes to define position and attitude of the user, in real time
navigation using images and photogrammetric algorithms, and
eventually inertial sensors (De Agostino et al., 2010).
The IBN can be realized in different ways, in particular in this
research the authors have following these steps (Figure 1):
it is initialize through a first photo that represent the
starting point for the localization;
then the navigation go on using the inertial sensors
that indicate to the user the direction and attitude to
follow;
after a specific time a new image is needed in order to
correct the inertial drift.
Figure 1 - The Image-Based Navigation procedure
To explain, the IBN has been divided in two phases:
image-based localization (IBL);
inertial sensors navigation (ISN).
The IBL is based on the matching between each photo and a
reference image extracted from a database of 3D images. As it
is better described in chapter 4, the IDB is defined by solid
images (Bornaz et al., 2003, Forno et al, 2013) that defines a 3D
information of the position. The IBL procedure adopts well-
known algorithms for image matching, such as SIFT and
RANSAC. In particular, the method is realized as described in
the following:
A. the matching between the two images is realized in
this way:
common features are extracted between the real time
image and the reference one using the SIFT algorithm
(Li et al., 2011);
key points are matched;
the estimation of the fundamental matrix using
RANSAC, in order to detect other outliers which were
not previously detected.
Image Based Navigation approach can help to estimate and
correct the inertial sensors drift, in order to improve the quality
of positioning up to 0.4 m. (Lingua et al. 2014) (Piras et al.,
2014).
After that, the reference image has been directly extracted from
the 3D IDB. The second step is to using this image to estimate
the 3D position:
B. parameters estimation:
the common features are translate in 3D information
using the related solid image
11 DLT parameters are estimated using the common points detected trough the features and are decoded to obtain the exterior orientation parameters in the first approximation;
using the collinearity equations, the external orientation parameters are defined.
the navigation solution (attitude, position) has been estimated.
Concerning the ISN procedure, it starts with the analysis of the
raw data of inertial sensors (acceleration, angular velocity and
magnitude of magnetic field), that can be directly registered
from the smartphone using “ad hoc” Android APPs. First of all,
it is necessary to filter these data, as it is better described in the
next part, in order to analyse the noise. Then the navigation
solution is extracted using a dedicated software developed at the
Politecnico di Torino and written in MATLAB® languages
This software was created to integrate GNSS and IMU
solutions, but now it is adapted in order to introduce the image
solution.
3. DESCRIPTION OF THE USED SENSORS
The IBN procedure can be adopted also for mass-market
devices, in particular the most popular sensors today available
are smartphone that now include many sensors useful for
Geomatics applications.
In this paper, the performance of the most famous devices
available off the shelf have been tested: Samsung Galaxy S4
Advanced, Samsung Galaxy S5 and IPhone 4. They have
embedded different internal sensors, such as digital camera and
GNSS receivers, even inertial platform based on gyroscopes,
accelerometers and magnetometers and RFID system for
smartphone devices. The technical characteristics of each one
are described in (Table 1).
These devices include sensors whose characteristics must be
known in order to realize a good positioning. In particular, it is
fundamental to characterize the noise level of the inertial
sensors and calibrate the camera, with purpose to remove the
lens distortions, which is fundamental for realizing a positioning
with a photogrammetric approach.
3.1 Sensors calibration
For the IBN we can take into account the errors due to:
camera lens distortions;
inertial sensor distortions.
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-3, 2014ISPRS Technical Commission III Symposium, 5 – 7 September 2014, Zurich, Switzerland
This contribution has been peer-reviewed.doi:10.5194/isprsarchives-XL-3-9-2014 10
The digital camera embedded in the smartphone are not-metric
sensors, so they require a calibration through an analytical
procedure in order to study their characteristics and to know the
mechanical-digital distortion parameters.
Name Samsung
Galaxy S5
Samsung
Galaxy S4
Advanced
IPhone 4
Cost 500€ 200€ 400€
OS
Android 4.4.2
TouchWiz UI
KitKat
Android 4.1.2 Jelly Bean
iOS 7.0.6
CPU Adreno 330 ARM Cortex-A9 dual-core
Apple4 - 800MHz
Digital
camera
Resolution
16Mpx 5Mpx 5Mpx
Type of
lens CMOS CMOS CMOS
A-GPS Yes Yes Yes
GNSS
receiver U-blox 6N U-blox 6N
Broadcom - BCM4750
Inertial
platform Yes Yes Yes
Table 1 - Devices and their principal characteristics
As well known, the optical system is composed by a set of
lenses with different curve shape. Lenses are pieces of glass
conveniently burnished having a spherical surface; the centre of
curvature of each portion of sphere is located on a straight line
also called lens optical axis. Thanks to its spherical shape, it is
possible to deviate ray light flowing through lenses.
A real photogrammetric lens has significant differences with
respect to the ideal one because:
the lens assembly misalignment;
the photogrammetric reference axis will not be the
optical axis, but a principal calibrated that, in the object
space, is perpendicular to the image plane;
the refractive and incidence angles do not match;
the main distance is slightly different from the main
optical distance;
the image plane is not perfectly perpendicular to the
optical axis.
At the end, the principal distance variation (Δc) is calculated,
trying to lead the mean value to zero.
These distortion curves could be represented using an odd-
degree polynomial function in ρ (distance from principal point
position inside image plane) :
(1)
( ) (
) (2)
( ) (
) (3)
In some sort of cameras, especially amateur one, it should be
considered even tangential distortion:
[ ( ) ](
) (4)
[ ( ) ](
) (5)
In the other hand, sensor is made of silicon wafer that is
substantially static in terms of geometry. However there exist a
distortion effect related to the geometry of the sensor: in theory
a pixel should be a perfect square and the rows matrix should be
perpendicular to the columns; in reality this does not happen.
Such distortion is time constant, depends only on sensor
construction, and is conveyed by a particular affine
transformation:
( ) ( ) (6)
So, using mass market devices is crucial to calibrate the lens in
order to know and model these parameters (Aicardi et al. 2014).
In particular, in this case, the radial distortions are only
considered because it has the most impact in the image
distortion. Tangential distortions have not been considered,
because the effects are not appreciable.
The analytical calibration mode of the cameras are usually
divided into on-the-job calibration and self-calibration, which
are based on the solution of the calculation of a bundle-
adjustment performed considering as unknowns the six external
orientation parameters of the images and the six parameters of
the camera calibration (ξ0, η0, c, K1, K2, K3) (Kraus, 1997).
Figure 2 - Calibration field
In this application, the self-calibration procedure has been
used. This is based on the determination of the calibration
parameters carried out independently by the procedures of the
photogrammetric survey. This method is usually performed by
preparing a calibration grid, specifically made, in which the
coordinates of the target are known with extreme precision.
The calibration has been realized using a dedicated calibration
field, which is externally materialized in the Geomatics
Laboratory at the Politecnico di Torino (Figure 2).
The software Leica Photogrammetric Suite (LPS) by ERDAS
has been used for the self-calibration of the device. For all
devices, the distortions have been estimated and they are
reported in Figure 3, Figure 4 and Figure 5.
As it is shown in Table 2, the Samsung S4 is the smartphone
with minor radial distortion that, correcting it by a linear trend,
can be neglected. The others two devices have a similar
distortion pattern which is about twice then the first one.
Another aspect that has to be considered is the stability of the
internal inertial sensors, whose performances are usually not
declared. So, we performed a 6 hours static test in order to
acquire the raw data (angular velocity and accelerations) of each
Smartphone and we analyse the stability.
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-3, 2014ISPRS Technical Commission III Symposium, 5 – 7 September 2014, Zurich, Switzerland
This contribution has been peer-reviewed.doi:10.5194/isprsarchives-XL-3-9-2014 11
Figure 3 - Radial lens distortion resulting from Samsung S4
calibration
Figure 4 - Radial lens distortion resulting from Samsung S5
calibration
Figure 5 - Radial lens distortion resulting from IPhone4
calibration
The analysed parameters are related to the acceleration and
gyroscope, and also to the attitude, which is real time calculated
from the inertial platform and could be used as the “a priori”
attitude for the IBN.Figure 6, Figure 7 and Figure 8 show the
result of the Samsung S4 inertial platform analysis. Same
analysis has been conducted for the Samsung S5, obtaining the
following results (Figure 9, Figure 10, Figure 11).
[µm] Samsung S4 Samsung S5 IPhone4
max 10,24 35,74 33,57
resmax 3,58 7,20 6,59
Table 2 - Devices radial distortion
Figure 6 - Acceleration residuals for Samsung S4
Figure 7 - Gyroscope residuals for Samsung S4
Figure 8 - Roll, Pitch and Yaw stability for Samsung S4
As it is possible to see in the next graphs, it seems that the
accelerometers of Samsung Galaxy S5 are better than the
Samsung S4 ones, while the gyros of S5 are worse than S4.
Also the attitude is quite different: while in the Samsung S4 the
Roll component is quite noisier, in the Galaxy S5 both roll and
yaw components are very stable.
The stability and the performances of the IPhone4 are described
in Figure 12, Figure 13 and Figure 14.
y = 3,4824x
R2 = 0,8118-6,00
-4,00
-2,00
0,00
2,00
4,00
6,00
8,00
10,00
12,00
0 0,5 1 1,5 2 2,5 3
r [mm]
Dr
[mic
ron
]
Dr
Dr residual
Trend
y = 8,179x
R2 = 0,8708
-10,00
-5,00
0,00
5,00
10,00
15,00
20,00
25,00
30,00
35,00
40,00
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5
r [mm]
Dr
[mic
ron
]
Dr
Dr residual
Lineare (Dr)
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-3, 2014ISPRS Technical Commission III Symposium, 5 – 7 September 2014, Zurich, Switzerland
This contribution has been peer-reviewed.doi:10.5194/isprsarchives-XL-3-9-2014 12
Figure 9- Acceleration residuals for Samsung Galaxy S5
Figure 10 - Gyroscope residuals for Samsung Galaxy S5
Figure 11 - Roll, Pitch and Yaw stability for Samsung S5
Starting from Table 3 it is possible to note that in general the
accelerometers of Samsung S5 are slightly better than the S4
ones while it is the opposite for the gyro components.
In general it is possible to affirm that the iPhone4
accelerometers are the best (if these three smartphones are
compared) while the gyros are the worst. It must to be
underlined that in this paper we don’t want to determine which
smartphone is the “best” but which of them is more useful for
Image Based Navigation purposes.
Figure 12 - Acceleration residuals for IPhone4
Figure 13 - Gyroscope residuals for IPhone4
Figure 14 - Roll, Pitch and Yaw stability for IPhone4
For the inertial sensors, it is possible to try to correct the noise
of the data making a filtering. To make this it is possible to use
the wavelet, which are signal representations by the use of a
waveform oscillating. There are different kinds of these
representations, but, for this case, the Daubechies wavelets were
chosen to correct the data and, in particular, we used a
Daubechies4 at a Level7.
In Figure 15, it is possible to see the wavelet de-noising
approach thanks to the Matlab toolbox while in Figure 16 it is
reported an example of the signal after the wavelet filtering (in
red it is possible to see the original signal while in black the de-
noised one).
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-3, 2014ISPRS Technical Commission III Symposium, 5 – 7 September 2014, Zurich, Switzerland
This contribution has been peer-reviewed.doi:10.5194/isprsarchives-XL-3-9-2014 13
RMS X Y Z
Samsung
S4
Acceler. [g] 0.0066 0.0055 0.0057
Gyro. [rad/s] 0.0008 0.0017 0.0018
Samsung
S5
Acceler. [g] 0.0046 0.0061 0.0002
Gyro. [rad/s] 0.0017 0.0028 0.0009
IPhone4 Acceler. [g] 0.0028 0.0024 0.0042
Gyro. [rad/s] 0.0048 0.0032 0.0043
Table 3 - RMS of the inertial sensors
Figure 15 - Wavelet denoising
Figure 16 - Original (red) and de-noised (black) signals
4. 3D SOLID IMAGE DATABASE GENERATION
In IBN, the position of the image is estimated from the
comparison between the image taken from the user and the
reference image extract from a 3D IDB, that in this case is
defined by means solid images.
A solid image is a synthetic image in which for each pixel is
associated the information on the relative position of the spatial
point projected on the image, expressed in 3D coordinates in a
defined reference system, considering the camera parameters.
These images are extracted from a 3D model that can be
generated in different ways: using existing 3D City Models,
making a terrestrial or an aerial survey or collecting the data
trough Mobile Mapping Systems.
For this application, a terrestrial LiDAR survey has been used,
that also allows the acquisition of images using an integrated
camera, with purpose to obtain a coloured points cloud.
Five different scans have been acquired and mounted in a single
model in a common reference system. As result of the process, a
geo-referenced coloured point cloud of the environment is
provided, on which you can directly read 3D coordinates and
colour of the points of interest: this model is used for the
extraction of the images, and complementary spatial
information.
Figure 17 - Examples of the image DB
Now, the solid image can be automatically generated by means
of these steps (Lingua et al. 2014):
an empty solid image (RGB and range) is generated using the number of pixels in column and row of the
solid image (ncol, nrow);
a subset of colored points (Xi, Yi, Zi) with i=1:n,
(n=number of selected points) can be extracted from the
original RGB point cloud according to a selection
volume that can be defined by a sector of a sphere with:
- center in the location of generated solid image;
- axis direction coincident with the optical axis of synthetic solid image;
- radius R;
- amplitude defined by an angle ( 90°) that is half the cone angle measured from direction axis;
for each selected colored point, a distance di respect the
location of generated solid image is calculated:
20
2
0
2
0 ZZYYXXd iiii
each selected RGB point is projected on the solid image defining its image coordinates ( ) by means of the internal and external orientation parameters inside the collinearity equations:
the image coordinates ( ) are converted in pixel coordinates ( ) using:
22
row
pox
ii
col
pix
ii
n
dr
n
dc
(4)
the RGB values of each point are wrote inside the cell of image RGB matrices in the position ( );
the distance value di is wrote inside the cell of range
image matrix in the position ( )( ); at the end of the procedure, pixels still void are filled by
means of an interpolation algorithm based on nearest filled pixels.
The DB can be create under different conditions, for example it is possible to set, the camera parameters, the position of the hold centre, the grip axis and the number of images for each point. At
the end of the procedure, we obtain a set of images (Errore. L'origine riferimento non è stata trovata.) with the information about position and attitude. These images can be used to initialize the IBN procedure, making the first photo of the site, and to correct the navigation using a new image.
5. TEST AND CASE STUDIES
The tests have been realized in a courtyard in our campus
(Figure 18). The track (red line in Figure 18) was especially
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-3, 2014ISPRS Technical Commission III Symposium, 5 – 7 September 2014, Zurich, Switzerland
This contribution has been peer-reviewed.doi:10.5194/isprsarchives-XL-3-9-2014 14
performed in an area that present many windows and with an
high repeatability of the modules.
This area has been chosen to simulate urban canyon or indoor
location (the GNSS data has not been acquired/used) in a
convenient situation (all the area is visible from a unique total
station position) with typical noise in indoor image acquisitions
(walking people, variable condition of light, shadows,…).
In this area, the test has been performed walking on the same
path using the three different smartphones (a) mounted on a
special support (Figure 19), realized by our Geomatics
Laboratory, which allow to support:
an inertial platform IMU-MEMS Microstrain
3DMGX35 (b) with external antenna (c);
a 360° retroflector (d).
Figure 18 - Test site and track
During the tests each smartphone sensor has recorded own
inertial sensors data using a dedicated Android App, called
“AndroSensor”, that gives graphical information and text (.csv)
output.
The reference trajectory has been defined tracking the
smartphone position in continuous with a total station, through
the retroflector; in this way, the position of the support has been
measured with an accuracy of few mm..
Moreover, the data from an external IMU (microstrain) has been
stored using a computer, with purpose to have reference values
of the attitude.
For this application different types of data were acquired from
the devices, in particular:
Samsung S4: images
Samsung S5: videos
IPhone4: images and videos.
After the test, all of IMU data files concerning each sensor, the
image/video of the tracks and the reference data for the
comparison between the estimated and the real solution have
been available and they have been processed.
6. FIRST RESULTS
First results highlight how the IBN improve the correctness of
positioning in indoor application, in particular describe the
benefit of this technique with respect the navigation IMU only
or, worst, with GNSS only. A comparison of the acquired tracks
with all smartphones involved in our test and the IBN solution
has been realized. The result are shown in Figure 20, where the
green line is the solution estimate with MEMS platform and
using an integrated solution GNSS-IMU, where the GNSS data
was available
Figure 19 - The system used to acquire the data
The result are shown in Figure 20, where the green line is the
solution estimate with MEMS platform and using an integrated
solution GNSS-IMU, where the GNSS data was available.
Figure 20 - Track results comparison
The IBN solution obtained with IPHONE sensors (red line in
Figure 20) brings to reach a horizontal error loop equal to 1.4 m
and a vertical error loop equal to 0.26 m, considering a session
length amount to 6 minutes. The error in the angular estimation
is about 1.8 gons, estimating a noise equal to 0.4 gons.
The IBN solution obtained with Samsung S4 sensors (blue line
in Figure 20) brings to reach a horizontal error loop equal to 1.6
m and a vertical error loop equal to 0.33 m, considering a
session length amount to 6 minutes. The error in the angular
estimation is about 1.6 gons, estimating a noise equal to 0.34
gons.
Finally, considering the Samsung Galaxy S5 sensors (orange
line in Figure 20) the horizontal error loop obtained is equal to
1.5 m and a vertical error loop equal to 0.38 m, considering a
session length amount to 6 minutes. The error in the angular
estimation is about 1.4 gons, estimating a noise equal to 0.29
gons.
7. CONCLUSIONS
The realized tests highlight that the modern smartphone can
really help the user to define its position even if the actual
application and technology are not developed to take the greater
advantages from the internal sensors, in fact it is quite difficult
to recording the raw data or to have the direct access to the
internal sensor.
The performance improvement, in term of precision of position,
could be obtained making a calibration of the lens and adopting
specific algorithms for data analysis, in particular for outlier
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-3, 2014ISPRS Technical Commission III Symposium, 5 – 7 September 2014, Zurich, Switzerland
This contribution has been peer-reviewed.doi:10.5194/isprsarchives-XL-3-9-2014 15
detection in the images and to filtering the IMU data. It is
generally possible to affirm that the “best” solution in term of
horizontal and vertical error loop is obtained with iPhone
smartphone, while considering the angular estimation the
Samsung Galaxy S5 provide the best results.
The accuracy of the IBN positioning depends on the definition
of the three dimensional, which can be built in different way.
Furthermore, working with smartphones technologies, it is
important to use adequate algorithm for compressing the
images, in order to send it to the server where the IDB has been
stored.
These first tests have demonstrated the feasibility and the
performances of the IBN with modern smartphones: in the
future this approach will be investigated deeply in order to
obtain better results also useful for indoor positioning.
8. ACNOWLEGMENTS
Authors would like to thank the prof. Bendea, who has realized
the “butterfly” support.
A part of the research presented in the paper refers to part of the
activities carried out in the Pro-Vision Project (Title: Sistema
Prototipale per le Verifiche di Visibilità su Infrastrutture di
Trasporto Esistenti in Ambito di Smart City), which is funded
by the Regione Piemonte (F.E.S.R. 2007/2013).
Other parts of this research have been financed from
TELECOM Italia by means of the project: “"Navigazione in
tempo reale mediante smartphone in ambienti indoor e outdoor
basandosi su tecniche di localizzazione per immagini".
9. REFERENCES
Afzal M.H., Renaudin V., Lachapelle G., 2011. Use of Earth’s
Magnetic Field for Mitigating Gyroscope Errors Regardless of
Magnetic Perturbation. Published in Sensors
Aicardi I., Lingua A.M., Piras M., 2014. Evaluation of mass
market devices for the documentation of the cultural heritage. In
2014 ISPRS Conference, Commission V. June 23-25, Riva del
Garda.
Bornaz L., Dequal S. ,2003. A new concept: the solid image.
CIPA 2003 Proceedings of XIXth International Symposium:
169-174.
De Agostino M., Lingua A., Nex F., Piras M. 2010,
GIMPhI: a novel vision-based navigation approach for low cost
MMS. In: Position Location and Navigation Symposium
(PLANS), 2010 IEEE/ION, Indian Wells, CA, USA, 4-6 May
2010. pp. 1238-1244
Frank K., Munoz Diaz E., Robertson P., Fuentes Sanchez F.J.,
2014. Bayesian Recognition of Safety Relevant Motion
Activities with Inertial Sensors and Barometer. In 2014 Plans
Conference proceedings. pp.174-184
Forno M.G., Lingua A., Lo Russo S., Taddia G., Piras M., 2013.
GSTOP: a new tool for 3D geomorphological survey and
mapping , EUROPEAN JOURNAL OF REMOTE SENSING,
Davide Travaglini, pp. 16, , Vol. 46, ISSN: 2279-7254, DOI:
10.5721/EuJRS20134613
Hatami A., Pahlavan K., 2005. A Comparative Performance
Evaluation of RSS-Based Positioning Algorithms Used in
WLAN Networks. IEEE Wireless Communications and
Networking Conference.
Kalliola, K., 2011. High Accuracy Indoor Positioning Based on
BLE. Nokia Research Center Presentation.
Kraus K., 1997. Photogrammetry II, Advanced methods and
applications. Ummler/ Bonn, ISBN 3-427-78694-3.
Kunze K., Lukowicz P., Partridge K., Begole B., 2009. Which
way am I facing: inferring horizontal device orientation from an
accelerometer signal. In International Symposium on Wearable
computers.
Lee S., Koo B., Jin M., Park C., Lee M.J., Kim S., 2014. Range-
Free Indoor Positioning System Using Smartphone with
Bluetooth Capability. In 2014 Plans Conference proceedings.
pp.657-662
Li X., Wang J., Li R., Ding W., 2011. Image-based positioning
with the use of geo-referenced SIFT features. Proceedings of the
Incorporating the International Symposium on GPS/GNSS
(IGNSS 2011), Sydney, Australia.
Lingua A.M., Aicardi I., Ghinamo G., Corbi C., Francini G.,
Lepsoy S., Lovisolo P., 2014. Technique based on 3D LIDAR
scanning and MPEG7 Visual Search Solutions. In 2014 ION
Conference, in press.
Piras M., Dabove P., Lingua A.M., Aicardi I., 2014. Indoor
Navigation Using Smartphone Technology: A Future Challenge
Or An Actual Possibility?. In 2014 Plans Conference
proceedings. pp.1343-1352
Rousseeuw J.P., Leroy A.M., 1987. Robust regression and
outlier detection. Wiley & Sons.
Schatzberg U., Banin L., Amizur Y., 2014. Enhanced WiFi ToF
indoor positioning system with MEMs-based INS and
pedometric information. Intel Corporation, Israel. In 2014 Plans
Conference proceedings. pp.185-192
Shin B., Lee S., Kim C., Kim J., Lee T., Kee C., Heo S., Rhee
H., 2014. Implementation and Performance Analysis of
Smartphone-based 3D PDR System with Hybrid Motion and
Heading Classifier. In 2014 Plans Conference proceedings.
pp.201-204
Werner M., Schauer L., Scharf A., 2014. Reliable Trajectory
Classification Using Wi-Fi Signal Strength in Indoor Scenarios.
In 2014 Plans Conference proceedings. pp.663-670
Woodman, O., Harle R., 2008. Pedestrian Localisation for
Indoor Environments. Proc. UbiComp, Seoul, Korea.
Woodman O.J., 2007. An introduction to inertial navigation.
Technical report
Yuan X., Liu C., Zhang S., Yu S., 2014. Indoor Pedestrian
Navigation Using Miniaturized Low-Cost MEMS Inertial
Measurement Units. In 2014 Plans Conference proceedings.
pp.487-492
Yunye J., Hong-Song T., Wee-Seng S., Wai-Choong W., 2011.
A Robust Dead-Reckoning Pedestrian Tracking System with
Low Cost Sensors. In 2011 IEEE International Conference on
Pervasive Computing and Communications
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-3, 2014ISPRS Technical Commission III Symposium, 5 – 7 September 2014, Zurich, Switzerland
This contribution has been peer-reviewed.doi:10.5194/isprsarchives-XL-3-9-2014 16